Women with dense areas and calcifications, observed on mammograms, have a higher risk of breast cancer. High-risk breast density patterns and the percentage of the breast which is dense have been associated with factors related to the risk of breast cancer among women without breast cancer (age, menopausal status, age at first live birth, parity, body mass index, hormone replacement therapy, alcohol use, frequency of exercise, family history of breast cancer and others). We have the opportunity to investigate whether mammographic densities and calcifications are in the pathway by which these factors are related to breast cancer. A combined data set, from four studies performed in western Washington state, contains 547 breast cancer cases (ascertained from a SEER registry) and 472 controls (ascertained by random digit dialing). Initial mammograms performed, before age 50, on study subjects were evaluated by a reference radiologist by Wolfe's classification (for breast density), percent breast density, and characteristics of calcifications. Previous analyses from this study have shown associations between the aforementioned mammographic characteristics and breast cancer (Thomas et al., 2002). We intend to extend analysis of this data to determine whether controlling for these mammographic features alters the association which factors that have been associated with breast cancer (ex. parity) have with breast cancer. The extent of this alteration of association will be evaluated through multivariate logistic regression models and techniques previously used in clinical trials to assess the value of surrogate endpoints. In addition, nested models, utilizing path modeling and mediator analysis will be used to establish pathways by which these factors associated with breast cancer may cause breast cancer, considering the possible mediating effect of mammographic patterns.